G’day everyone — building designer here, 18 years of residential work in Perth, Australia.
I spent most of those years watching rendering eat the hours that design deserved — so I built the tool I always wanted, and recently released the SketchUp extension for it.
What it does: one click captures your current SketchUp view and opens it in Silpu Studio+ — about a minute later it’s a photoreal image. No exporting, no material setup, no render farm. Reframe your view, hit re-capture, render again.
What makes it different from the AI renderers you’ve probably tried:
It respects your model. Your roof form, openings and materials render as you built them — unusual features are treated as deliberate design, not “corrected.”
Mark-up directing — circle a wall, drop a pin, write “stone here” — and it follows your instructions like a brief.
Real lighting — place fixtures with actual IES photometric profiles, then relight day→dusk→night without re-rendering.
It dresses the world — model just the building; the street, neighbours, garden and interior furnishing are added around it, true to your project’s location.
Honest note: it’s AI — beautiful concept imagery, not documentation. Always check results against your drawings.
I’m one person building this and I answer everything — if you throw a stubborn model at it and it misses, post the result here or email me. Stubborn models are exactly what I sharpen it on.
question: Why would I pay $600 a year for this service which has been made using an AI coding platform rather than one from a well known and well respected developer?
What advantages does this have other the big ones and the countless other AI render extensions?
I counted 16 em dashes in the OP’s post. Wonder who/what wrote it? Also, more seriously, the 5 minute getting started guide felt like it would take a good 30 mins to read fully. There are so many settings to control (a good thing?) that it feels like setting up a scene and rendering is V-Ray will take less effort.
Fair question Dave. It’s my own Classic perpetual license, from back when Trimble still sold them. 2021 was the last version my license covers, so that’s where I stayed when subscriptions came in. Eighteen years of residential work hasn’t needed anything 2021 can’t model.
The extension itself supports SketchUp 2017 and newer, so it runs happily on the current subscription versions too.
G’day Adam, very good questions, and from someone at Elmtec I’d expect exactly these.
On price: there is no subscription. No $600 a year. Silpu is pay as you go: a render costs about US$0.97, you start with free renders (no card), and credits never expire. Twenty renders in a month is roughly $19. If something on my site suggested $600 a year, tell me where, because that’s a bug and I’ll fix it today.
On being built with AI tools: guilty, and openly. I’m a building designer, 18 years of residential work in Perth, not a career software developer. AI helps me write code the same way CAD helps me draw. What it can’t provide is the judgement of what a render must get right for a client meeting. Now, when I ride the train from Kwinana to Perth Station, I can ask my draftee to send me a screen shot and render on my phone. That part is the 18 years.
On what’s different from the big ones: most AI renderers re-imagine your model. Nearly all of my engineering goes the other way: keeping the roof form, openings and materials exactly as modelled, and giving you director controls (mark-up on the model, a materials schedule, real IES photometric profiles) instead of a slot machine. On the Day to Night note: that one isn’t a single image, it bakes a scrubbable dawn-to-dusk-to-night sequence you can slide through and export.
But those are claims, and claims are cheap. The free renders exist so nobody has to take my word for anything. Throw your most stubborn model at it and judge the result. If it misses, post it here where everyone can see, and you’ll also see what I do about it. I got free animations you could use with music choices, and yes created by AI so no copyright infringements for users.
The em dashes: fair cop. I drafted the post with AI help, the same way I openly build with it. The tool and the years behind it are mine.
Your settings point is the most useful feedback in this thread, because the reality is the opposite of how it read: the required workflow is two clicks. Drop in a screenshot, press Generate, done. The defaults set the scene, light and materials. Everything else exists for the day a client says “make that wall red brick” and you want to be obeyed rather than surprised.
If the handbook made it feel like V-Ray-grade setup, that’s a writing failure on my end, so I’ve fixed it today: the handbook now opens with a section called “The two-click version”, right at the top of Lesson 01. Refresh and it’s there.
I’d genuinely value the second half of the test too: try the two-click path on one of your own models and tell me straight how it compares to your V-Ray setup time. As with most AI, Silpu renders with gray models, you just have to make sure “What Silpu read on the dropdown is as per your client’s material wishes”.
Ha, either way that cuts, the cure is the same: go look.
The free renders are there for exactly that. And a standing offer for the regulars in this thread: if you want more than a taste, email me at edwin@silpu.com and I’ll load your account with enough credits to test it properly. A sceptic putting it through its paces in public is worth more to me than any ad I could buy.
